When to use a carousel and when to use a single image ad on Facebook, with a decision table, benchmark medians and a small-budget test.
Use a carousel when you want to show several products, several features or a step-by-step story. Use a single image ad when you have one product or one message and want the simplest, fastest ad to make. Neither format wins everywhere, so the reliable answer is to test them on your own account with a stop rule set before you spend.
This guide gives you a decision table by goal and product count, what each format does well, a small-budget test you can copy and a way to read the results.
Creative Insights breaks down creatives to show what makes winners win.
The quick answer by goal
| Your situation | Start with | Why |
|---|---|---|
| One hero product, one offer | Single image | One clear picture and one message, nothing to swipe |
| 3 to 10 products from one collection | Carousel | Each card can link to its own landing page |
| Explaining how something works | Carousel | Cards walk through the steps in order |
| Telling a before-and-after or brand story | Carousel | Images and videos play in succession |
| Quick sale or discount announcement | Single image | The offer fits in one headline |
| Cold audience, no product knowledge | Single image first, carousel to test | Easier to read the result of one clean hook |
| Retargeting people who saw products | Carousel to test against single image | Showing several viewed products may help; test it |
These are starting points, not rules. Use the test below to settle it for your brand.
What each format does well
Carousel ads
Meta's carousel ads guide describes the format as 2 or more images and videos in a single ad, each with its own headline, description, link and call to action. Each one is a card, and people swipe on mobile or click arrows on a computer. Meta lists these uses:
- Feature multiple products that link to different landing pages, which Meta says can help raise clickthrough rate.
- Highlight multiple features of a single product.
- Explain a process, step by step.
- Tell a story, or create one long image across all cards.
The cost is production effort: every card needs its own image, headline and link, and a weak second card can lose the viewer. For a deeper walkthrough, see Facebook carousel ads.
Single image ads
A single image ad shows one picture, one headline and one call to action. It does one thing: put one idea in front of someone quickly. It is cheaper to produce, easier to run many versions of and easier to read in a test, because there is one image and one message to judge. The cost is that it can show only one product or one idea.
Decision table: products, goal, audience
| Question | If yes, lean toward |
|---|---|
| Do you have more than one product to show? | Carousel, one product per card |
| Does the offer fit in one sentence? | Single image |
| Does the buyer need to see how it works? | Carousel, one step per card |
| Do you make new creative every week? | Single image, easier to produce in volume |
| Do you want traffic to different pages? | Carousel, one link per card |
| Is the audience cold? | Single image first, so the hook is easy to read |
Single image and carousel are two of several formats. See ad creative formats for Meta for the full list before you commit.
Costs and results: benchmark ranges
The benchmarks below are not split by ad format, so treat any single "carousel beats single image by X%" number with caution. What you can anchor on is the cost of reaching and clicking people on Meta overall. These medians come from Madgicx's Meta ads benchmarks, measured September 1 to 30, 2026, in USD.
| Metric (median) | All industries | Apparel and Fashion |
|---|---|---|
| Cost per click (all clicks) | $0.29 | $0.26 |
| Cost per 1,000 impressions (CPM) | $9.72 | $11.24 |
| Click-through rate (all clicks) | 2.97% | 3.96% |
| Cost per link click | $0.53 | $0.45 |
The overall median link click-through rate is 1.53%. The medians are the middle value, and the benchmark page also shows a wide middle 50% range: for all clicks, the overall cost per click runs from $0.06 to $0.85. Industry matters a lot, too: Home and Local Services has the highest median cost per click (all clicks) at $1.71, while Home Goods and Furnishings has the lowest at $0.05. Link-click and all-click figures use different click definitions, so compare like with like.
Use these as the reference for your account. If your single image ads sit near the median and your carousel does too, the format is not your problem; the offer or the audience is.
How to test with a small budget
A fair test changes only the format. Set it up like this:
- One campaign, one objective. Use the same goal for both ads, such as purchases.
- One ad set per format, or two ads in one ad set. Separate ad sets give you cleaner spend control. Two ads in one ad set let Meta's delivery decide, and it may send most of the spend to one ad before the other has a chance.
- Same audience, same placements, same days. Only the format differs.
- Same offer and same headline. The carousel's first card should carry the same idea as the single image.
- For carousels, use real card content. On Facebook a carousel needs between 2 and 10 cards; keep each card to one idea, and respect the copy limits in Meta's carousel ad specs (125 characters of text, 40 for the headline).
A sample budget and stop rule
No public source gives a safe test budget, so build one from your own numbers. Here is an example with a made-up target; swap in yours.
| Step | Example |
|---|---|
| Your target cost per purchase | $25 |
| Purchases you want each ad to have the chance to produce | 10 |
| Budget per ad | 10 x $25 = $250 |
| Two ads (one per format) | $500 in total |
| Test length | 14 days, about $36 a day |
| Early stop rule | Pause an ad that has spent 3 x your target cost ($75) with no purchase |
The numbers are an illustration of the method, not a benchmark. The method is what matters: price the test from your own target cost, decide the stop rule before launch and do not change the ads mid-test.
Reading the results
- Compare cost per purchase first. Click-through rate is useful, but a format can win clicks and lose sales.
- Look at the gap, not the order. If the two ads are close, call it a tie and test something bigger, such as the hook or the offer.
- Check what happened inside the carousel. If most people stop at the first card, the format is not the issue; the later cards are.
- Repeat before you commit. One test is a hint. A second test on another product tells you whether it holds.
- Compare against your benchmark. Use the medians above as a sanity check, and see Conversions API vs Meta Pixel if your purchase numbers look low or inconsistent, since missing tracking can make a winner look like a loser.
Mistakes that ruin a format test
- Changing the offer along with the format. If the carousel also has a discount, you are testing the discount.
- Judging too early. Early numbers can move while delivery is still learning about an ad. Wait until each ad has had its planned budget.
- Weak second and third cards. A carousel is only as good as the card people reach. Put your best card first and make every later card earn its place.
- Different audiences or dates. A format that ran on a payday weekend will beat one that ran midweek for reasons unrelated to the format.
- Stopping at one winner. Run a second product through the same test before you change how you build every ad.
- Ignoring tracking. If purchases are under-reported, both ads look worse than they are and the comparison gets noisy.
Find what makes the winning ad win with Madgicx
When a format test finishes, the question is why one ad won. Madgicx Creative Insights breaks down every creative to show what makes winning ads win, which helps you read a format test: was it the carousel, or the hook on its first card? Then take the winner into your next round. For a wider list of tools that help with ad results, see AI advertising tools to scale your ROI.
FAQ
Are carousel ads better than single image ads?
Not always. Carousels suit several products, features or a story. Single images suit one product and one message. Test both on your account and compare cost per purchase.
How many cards can a Facebook carousel have?
From 2 to 10 cards, according to Meta's carousel ad specs. Each card can have its own headline, description and link.
How much should I spend to test carousel vs single image?
Work it out from your target cost per purchase: multiply it by the number of purchases you want each ad to have a chance to produce. In the example above, $250 per ad at a $25 target.
When should I stop a losing ad in a format test?
Set the rule before launch, for example pausing an ad that has spent three times your target cost with no purchase. Do not stop an ad on its first few hours.
Which format is cheaper to make?
A single image is cheaper, because it needs one picture and one message. A carousel needs an image, headline and link for each card.
